Retinopathy of Prematurity (ROP) remains a serious public health problem causing significant visual sequelae in an estimated 2,060 premature survivors per year in the United States or 4,400 without the benefits of Cryotherapy. Cryotherapy is not always successful, ablates a significant portion of the retina, and its long term sequelae are unknown. Case control studies and anecdotal reports of treatment of ROP with oxygen as well as studies conducted in an animal model demonstrate that chronic (weeks-long) hypoxia during the ROP healing process may be a critical factor in determining which cases of ROP heal, an hypothesis supported by the association with retinal hypoxia in other proliferative retinopathies, such as diabetic retinopathy. Ironically, concerns over precipitating ROP by giving too much oxygen to infants with immature retinae has lead to strict oxygen management. Premature infants with unstable cardiopulmonary status and this minimal level of oxygen administration can experience episodic hypoxia leading to pulmonary hypertension and cor pulmonale. In practice, oxygen is given more liberally to these infants as soon as their retinae become fully vascularized, and thus free from the risk of ROP. These data justify testing of the hypothesis that "Supplemental Therapeutic Oxygen for Prethreshold ROP (STOP-ROP) will reduce by at least 1/3 the number of infants progressing to Threshold ROP." This proposal is the Lead Application for that test and is submitted with the Manual of Procedures, and proposals from the Coordinating and Participating Centers. Infants who develop Prethreshold ROP (as defined in the CRYO-ROP study) will receive continuous oxygen saturation monitoring, and randomly assigned oxygen administration at one of two specified oxygen levels, conventional (pulse oximetry of 89-94% saturation) vs supplemental (pulse oximetry of 96-99%). The primary outcome variable will be the proportion of infants who progress to Threshold ROP. The projected sample size of 880 is compatible with a 3 year enrollment in 15-30 centers; interim analyses will be conducted. Data on specific ophthalmic, neonatal and cost outcomes will be collected prospectively through 3 months corrected gestational age. Should this treatment prove to be both efficacious and safe, infants with moderate ROP prethreshold (nearly 5000 each year) would benefit from its use, both in their ophthalmic and neonatal outcomes.